ABSTRACT
We used staining of tissue sections by lectin conjugates to screen inbred strains of mice for polymorphisms which could be used as histological markers of chimaerism. We found one polymorphism, which involves reciprocal patterns of expression of binding sites for the N-acetyl-galactosamine-binding lectins from Dolichos biflorus (DBA), Helixpomatia (HPA) and Wisteria floribunda (WFA) on intestinal epithelium and vascular endothelium. The polymorphism is due to alleles at a single locus, designated Dlb-1 (for Dolichos lectin binding). Of 29 inbred strains examined, 3 are D1b-1a (type strain RIII-ro; gut epithelium - ve, vascular endothelium + ve), and 26 are D1b-1b (type strain C57BL/6J; gut epithelium + ve, vascular endothelium − ve). In RIII-ro and C57BL/6J embryos, the polymorphic difference is not clearly present until day 11 of gestation. Before then, embryos of both strains express binding sites on gut epithelium and on endothelium.
The temporal and tissue-specific patterns of expression of lectin-binding sites may result from differences in expression of an N-acetyl galactosaminosyl transferase. If so, elucidation of the genetic basis of the polymorphism might provide an insight into the mechanisms of developmental regulation of glycosyltransferase activity.
INTRODUCTION
We screened inbred mouse strains for histological markers of chimaerism by staining cryostat sections of tissues from different strains with a panel of peroxidase-conjugated lectins. By analogy with the blood group antigens in man (Kapadia, Feizi & Evans, 1981), we thought that polymorphisms between inbred mouse strains would most likely be carbohydrate based.
We found only one polymorphism (Ponder & Wilkinson, 1983). It is recognized by the lectins from Dolichos biflorus (DBA), Helix pomatia (HPA) and Wisteria floribunda (WFA), which have specificity for terminal non-reducing N-acetyl galactosamine residues (Etzler, 1972; Hammarstrom, 1972; Kurokawa, Tsuda & Sugino, 1976). The polymorphism is confined to intestinal epithelium and vascular endothelium, but in those tissues it can be used as a marker for studying clonal development in aggregation chimaeras (Schmidt, Garbutt, Wilkinson & Ponder, in press; Ponder et al. 1985). The polymorphism is unusual in that the pattern of expression of lectin-binding sites is reciprocal in the two polymorphic types. In this paper we describe the appearance of the polymorphism during embryonic development, the reciprocal pattern in adult mice, and genetic data which indicate that the reciprocal patterns are determined by alleles at a single locus.
MATERIALS AND METHODS
Mice
CBA/Ca mice were bred in the animal house of the Institute of Cancer Research. DDK, RHI-ro, C57BL/6J, BIO.A., and C57BL/10ScSn, were bred at the MRC Laboratories, Carshalton. Other mice of the inbred strains shown in Tables 1 and 2 were obtained from Olac (Bicester, UK) or from the Medical Research Council, Mill Hill, London and housed at the MRC Laboratories, Carshalton.
Male and female mice, 4 to 24 weeks old, were used for the screening for polymorphisms and in the study of the strain distribution of DBA binding. Embryos studied were from DDK, RHI-ro, and C57BL/6J strains, raised as above. Fl, backcross and F2 animals were bred at the MRC Laboratories, Carshalton.
Lectins, lectin conjugates and antibodies
Peroxidase conjugates of Ricinus communis agglutinin I and of lectins from Mangifera indica, Cotoneaster and Jack fruit were a gift from Dr J. A. Forrester (Institute of Cancer Research). Other lectins were purchased as purified lectin or as peroxidase or fluorescein isothiocyanate (FITC) conjugate from Sigma, Poole, England. Peroxidase conjugates of Dolichos biflorus agglutinin (DBA) (Sigma No LI 135) were prepared in our laboratory by the periodate method as previously described (Ponder & Wilkinson, 1983). Alkaline phosphatase conjugates of purified lectins from Helix pomatia, Wisteria floribunda, Vicia villosa and Codium fragile were prepared by glutaraldehyde conjugation (Ponder & Wilkinson, 1983) using alkaline phosphatase from calf intestine (Boehringer, No 567744). Rabbit antiserum to Forssman antigen (Willison et al. 1982) was a gift from Dr K. Willison (Institute of Cancer Research). Sugars were obtained from Sigma.
Histological methods
Methods were as described by Ponder & Wilkinson (1983). The initial screening of mouse tissues for polymorphisms was carried out on cryostat sections fixed in 10 % formol saline after cutting. Subsequent studies of the distribution of DBA, HPA and WFA-binding sites in embryonic and adult tissues were made on methacarn fixed, paraffin-embedded tissue (except that tissue from F1 and F2 animals used in the genetic studies and tissues used for enzyme digestions were fixed in 10 % formol saline: see below). Of the fixatives tried (10 % formol saline pH 7,0·1 % to 2% buffered glutaraldehyde, benzoquinone, 4°C acetone, 4°C absolute ethanol, Carnoy’s, Bouin’s, and methacarn), only glutaraldehyde above 0·2 % concentration caused loss of DBA binding; but the cleanest and highest density staining was obtained with methacarn. Controls for the specificity of lectin staining were provided in each experiment by the inclusion of slides incubated without lectin conjugate, with lectin conjugate diluted in buffer containing the appropriate inhibiting sugar (final concentration 2 % w/v), and (for DBA) known positive and negative tissues.
Binding of the rabbit anti-Forssman antibody was demonstrated using a goat anti-rabbit alkaline phosphatase conjugate (Sigma, A8025).
Chemistry of the DBA-binding site
For enzyme digestions, cryostat sections fixed in formol saline for 10 min at room temperature, and formalin-fixed paraffin-embedded sections, were incubated as follows: (i) neuraminidase (Behring, Vibrio cholerae 1/d ml−1 ) diluted 1:10 in 0-2 M-acetate buffer pH 5·5 for 30 min at 37 °C. A positive control was provided by unmasking of sites for peanut lectin-peroxidase conjugates on similarly processed sections of human kidney, (ii) Trypsin (Sigma, type III from bovine pancreas) 1mg in 20 ml PBS pH7·5 for 15 min at 37°C: no control, (iii) Glycosidases (no controls): α-galactosidase (Sigma) 0·025μml−1 in 50 mM-acetate buffer pH 4·5 for 30min at room temperature; β-galactosidase (Sigma) 1 mg ml−1 in 50mM-Tris pH 7·5 at 37 °C for 15 min. Other treatments (on unfixed and formalin fixed cryostat sections): (i) NP40 (Nonidet P40) extraction: NP40 (Sigma) 1 % or 0·1 % in PBS pH7·5 for 15 min at room temperature; (ii) 10 % trichloroacetic acid (TCA): 2 min at 4 °C; (iii) chloroform/methanol: 2:1 chloroform: methanol for 2 min at room temperature followed by 1:2 chloroform: methanol for a further 2 min at room temperature. In each case, sections were washed in PBS after treatment and then incubated with DBA-peroxidase above.
Genetic studies
Mice were scored for Dlb-1 polymorphic type by DBA-peroxidase staining of vascular endothelium and intestinal epithelial cells in formalin-fixed paraffin-embedded sections of small intestine and adjacent mesentery.
RESULTS
Screening for polymorphisms
The mouse strains, lectins and tissues used in the initial screen for polymorphisms are shown in Table 1.
The screening revealed only one polymorphism, for binding of DBA-peroxidase to intestinal epithelium and vascular endothelium (Fig. 1). In RIII-ro and DDK mice, DBA-peroxidase bound to vascular endothelium but not to intestinal epithelium: in the other 8 strains, the reverse was true. The polymorphism was designated Dlb-1 for Dolichos lectin binding, RIII-ro being the type strain for D1b-1a, and C57BL/6J the type strain for Dlb-1b. Subsequently, (see below), the lectins from Helix pomatia (HPA) and from Wisteria floribunda (WFA), which like DBA have nominal specificity for terminal non-reducing N-acetyl galactosamine, were found to recognise the same polymorphism.
Polymorphism in expression of DBA-binding sites in adult small intestine. (A, E) C57BL/6J small intestine (A, × 160; E, detail of epithelium and lamina propria of villus × 400). (B) adjacent section to (A): sugar control (× 160). (C, F) RIII-ro small intestine (C, × 160; F, detail of epithelium and lamina propria of villus × 400). (D) adjacent section to (C): sugar control (× 160).
Methacarn fixed, paraffin-embedded 4μm sections of tissue from 8-week-old mice. Counterstained with haemalum. DBA-binding sites appear black, g, gut epithelium; l, lumen between villi; m, unidentified DBA-positive cells (see Table 4): p, Paneth cells; e, endothelium; Go, Golgi apparatus.
In C57BL/6J intestine, columnar, goblet and Paneth cells are DBA-positive, and vascular endothelium is DBA-negative. In RIII-ro intestine the reverse is true: the endothelium of arteries, veins and blood capillaries (but not lymphatics) is DBApositive; intestinal epithelium is DBA-negative.
Polymorphism in expression of DBA-binding sites in adult small intestine. (A, E) C57BL/6J small intestine (A, × 160; E, detail of epithelium and lamina propria of villus × 400). (B) adjacent section to (A): sugar control (× 160). (C, F) RIII-ro small intestine (C, × 160; F, detail of epithelium and lamina propria of villus × 400). (D) adjacent section to (C): sugar control (× 160).
Methacarn fixed, paraffin-embedded 4μm sections of tissue from 8-week-old mice. Counterstained with haemalum. DBA-binding sites appear black, g, gut epithelium; l, lumen between villi; m, unidentified DBA-positive cells (see Table 4): p, Paneth cells; e, endothelium; Go, Golgi apparatus.
In C57BL/6J intestine, columnar, goblet and Paneth cells are DBA-positive, and vascular endothelium is DBA-negative. In RIII-ro intestine the reverse is true: the endothelium of arteries, veins and blood capillaries (but not lymphatics) is DBApositive; intestinal epithelium is DBA-negative.
A further 19 strains and substrains have been examined for their D1b-1 type, the majority falling into the D1b-1b group (Table 2).
In order to assess the use of the polymorphism for studying clonal development in gut epithelium and vascular endothelium during embryogenesis, we also examined a series of embryos of C57BL/6J and RIII-ro strains at different ages (see below).
Distribution of DBA binding sites in RIII-ro and C57BL/6J tissues (Tables 3 and 4)
Polymorphic patterns of binding of DBA-peroxidase in tissue sections of RIII-YO and C57BLI6J mice

Non-polymorphic patterns of binding of DBA-peroxidase in tissue sections of RIII-ro and C57BLI6J mice

The distributions reported below were consistent for 20 adult mice of each strain (males and females, aged from 4 to 24 weeks), and four embryos of each strain at days 8, 9, 10, 11, 13 and 18 of gestation. DBA-peroxidase (rather than HPA or WFA) was used throughout these studies.
Adult mice
The polymorphism in DBA-binding sites in the adult was confined to intestinal epithelium and vascular endothelium (Fig. 1; Table 3). Non-polymorphic binding sites are listed in Table 4.
Notable departures from the general pattern of polymorphism in the adult mice were (i) that the colonic epithelium was DBA-negative distally in some Dlb-1b strains (CBA/Ca, C3H/Bi, BALB/c, AKR, DBA/2, but not C57BL/6J, C57B10, NZB, NZW, 129/RrJ) which were otherwise DBA-positive in intestinal epithelium; (ii) that in otherwise DBA-positive colonic epithelium, crypts adjacent to lymphoid follicles were often DBA-negative, but this was not so for crypts adjacent to lymphoid follicles in small intestine; and (iii) in adult Dlb-1a strain mice (endothelium-positive), DBA binding was not present on endothelium in certain tissues in which endothelial cells were DBA positive in 13-day embryos (see Ponder & Wilkinson, 1983).
Embryos
In C57BL/6J and RIII-ro embryos before 11 days gestation, the polymorphism in DBA-binding sites was not as clearly expressed as in the adult. Both endothelium and gut epithelium were stained with DBA-peroxidase. In each strain, however, the staining was stronger and more widely distributed on the tissue which would remain DBA-positive in the adult. Thus, (Figure 2) the DBA-peroxidase staining of gut epithelium in C57BL/6J embryos extended outside the basal lamina into the surrounding mesenchyme. Conversely, endothelial staining was widespread in RIII-ro embryos, but confined in C57BL/6J embryos to patchy staining in the large vessels. By 13 days of gestation, the adult polymorphic pattern was established, and it was maintained in 18-day embryos.
Polymorphism in expression of DBA binding sites in embryonic tissue. (A) C57BL/6J embryo: estimated days gestation. Transverse section through posterior body region (× 100). (B) RIII-ro embryo: estimated days gestation. Transverse section through posterior body region (× 100). Tissue processed as in Fig. 1. DBA-binding sites appear black, g, gut; a, dorsal aorta; e, endothelial cells; ys, yolk sac: In the C57BL/6J embryo there is strong DBA binding to gut epithelium and surrounding tissue, and scattered DBA-positive endothelial cells in large vessels. In the RIII-ro embryo, the gut epithelium is weakly DBA-positive (most of the staining is at the luminal surface), but endothelium in aorta and other vessels is strongly positive.
Polymorphism in expression of DBA binding sites in embryonic tissue. (A) C57BL/6J embryo: estimated days gestation. Transverse section through posterior body region (× 100). (B) RIII-ro embryo: estimated days gestation. Transverse section through posterior body region (× 100). Tissue processed as in Fig. 1. DBA-binding sites appear black, g, gut; a, dorsal aorta; e, endothelial cells; ys, yolk sac: In the C57BL/6J embryo there is strong DBA binding to gut epithelium and surrounding tissue, and scattered DBA-positive endothelial cells in large vessels. In the RIII-ro embryo, the gut epithelium is weakly DBA-positive (most of the staining is at the luminal surface), but endothelium in aorta and other vessels is strongly positive.
Chemistry of the DBA-binding site
Complete inhibition of DBA-peroxidase staining was obtained with 0 · 2 M-N-acetyl galactosamine, and slight reduction in staining intensity with 0 · 2 M-L-fucose. 0 · 2 M-D-glucose, D-galactose, D-mannose and N-acetyl glucosamine were without detectable effect.
To investigate whether DBA-binding sites were borne on glycoprotein or glycolipid, we extracted unfixed cryostat sections with chloroform-methanol (see Methods). The distribution and intensity of DBA-peroxidase staining was unaltered. To investigate whether masking by changes in membrane conformation (Willison et al. 1982) might be responsible for the lack of binding sites in DBAnegative tissues, we treated unfixed cryostat sections with 10 % cold trichloracetic acid (TCA) and 0-1% or 1%NP4O. 10%TCA and 0-l%NP40 were without effect. After 1%NP4O extraction staining was lost, but again without the appearance of DBA-positive areas in previously negative tissue. To investigate the possibility of masking of DBA-binding sites by sialic acid or by galactose residues [the sequence N-acetyl galactosamine-galactose is common in carbohydrate chains of glycoproteins (Beyer et al. 1981)], we digested tissue sections with neuraminidase and with α and β galactosidases. None of these altered the pattern of DBA-peroxidase staining.
Finally, as another approach to examine whether the DBA-positive/DBA-negative difference was due to addition or loss of a sugar, and if so, which, a further 10 lectins [from Helix pomatia (HPA) (Hammarstrom, 1972), Wisteria floribunda (WFA) (Kurokawa et al. 1976), Vicia villosa, Codium fragile, Sofora japónica and Jack fruit (all with specificity for N-acetyl galactosamine), Pisum sativum (α-D-glucose), Maclura pom í fera, Mangifera indica and Cotoneaster (α-D-galactose)] were examined for tissue-binding pattern. Only two of the lectins with nominal specificity for N-acetyl galactosamine residues, HPA and WFA, gave a pattern identical to DBA. The other lectins showed different patterns, and no polymorphism. Antibody to Forssman antigen, which contains a terminal N-acetyl galactosamine residue (Willison et al. 1982), also gave a different staining pattern, again without differences between C57BL/6J and RIII-ro intestinal epithelium or vascular endothelium.
Genetics of the polymorphism
Each of the 29 inbred strains and substrains (see Table 2) tested was either gutepithelium-positive, endothelium-negative (G+E−) or gut-epithelium-negative, endothelium-positive (G−E+). None was G+E+, and none G−E−. Similarly, of seven C57L × SWR recombinant inbred (RI) strains (generously provided by Dr B. Taylor, Jackson Laboratory) six were G+E− and one was G−E+.
All of 24 D1b-1a × D1b-F F1 animals (RIII-ro × DBA/2; RIII/ro × C57BL/6J; SWRXC57L; RIII/ro × CBA/CaLac) were G+E+. Of 36Fl×DIWb back-crosses, all were G+ and 15/36 were E+ ; and of 40 F1 × Dlb-1a backcrosses, all were E+ and 19/40 were G+. The ratio of G−×E+ : G+E+ : G+E_ phenotypes in 78F2 animals was 16:45:17.
DISCUSSION
Our description of the distribution of DBA-binding sites in embryonic and adult tissue amplifies those of Watanabe, Muramatsu, Shirane & Ugai, (1981) and Noguchi, Noguchi, Watanabe & Muramatsu (1982), who used only Dlb-1b strains. Although Van der Valk & Hageman (1982) used mice of Dlb-1b and Dlb-1b strains, they reported both intestinal epithelium and vascular endothelium to be DBA negative. The polymorphism which we now describe is of particular interest because of the reciprocal pattern in the two polymorphic types, which our results suggest is determined by alleles at a single locus.
The F1 and backcross results are those expected if the expression of DBA-binding sites in endothelium and in gut epithelium are each determined by a single dominant gene. The results suggest further that these genes are alleles at a single locus. Two unlinked loci would be expected to give four phenotypes: G+G+, G+E−, G”E+ and G−E−. In this case, the likelihood of finding in inbred strains only the two phenotypes which we have observed would be extremely small. Discounting substrains, we have examined 23 inbred strains and 7 RI strains. If the loci were unlinked and all phenotypes fully viable, the chance of observing only the G+E− and G−E+ phenotypes in these 30 strains would be (1/2)30 = < 10−9. Even if G−E− were lethal (we know from the F1 mice that G+E+ is viable), the chance would be only (2/3)30 = 5 × 10−6. The F2 results strengthen the argument. The 16:45:17 ratio of phenotypes in the F2 animals does not depart significantly (P = 0·66) from the 1:2:1 ratio expected for a single locus. While these results might also fit the 9:3:3 ratio that would be expected with two unlinked loci with dominance at each locus but with the double recessive homozygote (G−G−E−E−) being lethal, it would be very unusual for such a double recessive genotype to be lethal when neither G−G− nor E−E− is in itself lethal. We conclude that the Dlb-1a/Dlb-1b polymorphism is due to an allelic difference at a single locus.
The stability of the DB A-peroxidase staining to chloroform/methanol extraction suggests that some at least of the DBA-receptor carbohydrate is carried on glycoprotein. It is likely that the polymorphic differences and the changes in DBA-binding patterns during embryonic development are due to tissue-specific and temporal patterns of activity of glycosyltransferases (Hakamori, 1981; Kapadia et al. 1981). The failure of detergent, trypsin or trichloroacetic acid treatment of unfixed sections to influence the DBA-staining patterns suggests that they are not due to changes in cell membrane conformation which influence the accessibility of the DBA-binding site (Willison et al. 1982).
If differences in glycosyltransferase activity are indeed the basis for the polymorphism, DBA-negative tissues might either contain N-acetyl galactosamine (Gal-Nac) residues masked by addition of a further sugar residue, or they might lack Gal-Nac because of failure to transfer the sugar to its acceptor. The failure to unmask DBA-binding sites by enzyme digestion of tissue sections argues against masking as the basis of the DBA-binding polymorphism. Furthermore, one would expect masking to be dominant, which would produce a G−E− phenotpye in F1 tissues, but this was not found. It is most probable, therefore, that the polymorphism results from the presence or absence of tissue-specific expression of an N-acetyl galactosaminosyl transferase. This would be predicted to give the G+E+ phenotype in F1 animals which we have observed.
The D1b-1a strains DDK, SWR and RIII are all of European origin or derived from European stock, as are GRS/A, LIS/A, STS and MAS/A, which are also of D1b-1a type (G. Uiterdijk, personal communication). The D1b-1b strains were mostly derived in the USA from mice originally from China, through English fanciers and then to American and English laboratories (Festing, 1979). The reciprocal pattern of Dlb-1a/D1b-1b polymorphism may therefore have arisen from a single genetic event in an ancestor of one of these groups of strains. Identification of the genetic mechanism of the polymorphism might provide insight into the mechanisms of temporal and tissue-specific regulation of glycosyltransferase activity.
ACKNOWLEDGEMENTS
B. A. J. Ponder holds a Career Development Award from the Cancer Research Campaign. This work was supported by a programme grant to the Institute of Cancer Research jointly from the MRC and Cancer Research Campaign.
We thank Dr J. A. Forrester for help with the initial lectin screening, and Dr K. Willison for useful discussions.